GEE models are essentially regression equations, either linear or logistic regression, that allow for varying numbers of observations per participant, while controlling for autocorrelation (we used the AR1 structure; an exchangeable structure produced very similar results). Specifically, we used the GEE method to model the main effects of genotype, eBAC, and their interaction (at the subject-level) on each of the dependent variables of interest (i.e., vigor, negative mood, and urge to drink) while controlling for Begin Drink Report, as time-varying covariates. The GEE framework is most appropriate for this manuscript because we were interested in between-subject factors (i.e., genes) as predictors of differences in mean levels of an outcome (i.e., subjective response to alcohol and urge to drink) (Schwartz & Stone, 1998).
